Meituan-Robotics-0 / release.json
thingvellir's picture
Use complete VLA checkpoint; remove redundant Qwen weights
88267c7 verified
Raw History Blame Contribute Delete
809 Bytes
{
"display_name": "Meituan-Robotics-0",
"framework": "QwenMMDiT",
"checkpoint": "checkpoints/model.pt",
"original_checkpoint_basename": "steps_250000_pytorch_model.pt",
"training_step": 250000,
"action_dim": 20,
"action_horizon": 16,
"code_repository": "https://github.com/meituan/Meituan-Robotics-0",
"code_commit": "88c667933dfc941a71c9b142eb84b716ac4f0707",
"serialization": "PyTorch state_dict; unchanged from the selected source checkpoint",
"initialization": "Stage 1 FAST checkpoint at 430000 training steps",
"repo_id": "MeituanRobotics/Meituan-Robotics-0",
"checkpoint_contains": [
"Qwen3.5-4B backbone",
"MMDiT action head"
],
"backbone_metadata": "base_vlm/Qwen3.5-4B-Action/",
"separate_qwen_weight_file_required": false,
"checkpoint_tensor_count": 929
}